It also perpetuates them in another manner. Contrary to popular perception, DNA identification, like any other form of probabilistic identification, is fallible, and the possibility of false positives only increases with the size of the database. So, as Sheldon Krimsky and Tania Simoncelli point out in their analysis of US DNA data banks:
With all African American DNA on the database, it is reasonable to assume that false convictions in that population group will actually increase because DNA fallibility follows racial lines. In cases where individuals have been falsely convicted because of mishandled DNA, they have disproportionately been people of colour.16
Therefore, as Virginia Eubanks observes, ‘when automated decision-making tools are not built to explicitly dismantle structural inequities, their speed and scale intensify them’.17 This speaks not only to the issue of intensifying pre-existing institutional bias through ignoring it, but also to the third way in which transformative equality is particularly concerned with issues of technology: their disproportionate effect upon the vulnerable and the marginalized. To those whose fingerprints did not match while attempting to open a bank account, or who temporarily found themselves shut out of the online income-tax portal because of a mismatch of names between Aadhaar and their PAN cards, biometric authentication was a source of harassment, nuisance, and bother. However, for those who depended upon the public distribution system for their monthly supply of food, an authentication failure meant death.18
Just like the ‘false positives’ in the DNA database would disproportionately hit the most vulnerable in the context of criminal trials, Aadhaar’s technological regime, by design, impacted those most in need of their basic entitlements. Indeed, before even the stage of biometric authentication failure, the Aadhaar system shut out an entire class of people from accessing the benefits that it had been made the drawbridge to: homeless people, who had no address proof, and could therefore not even apply for an Aadhaar number.19
It is this manner of institutional disadvantage—in its diverse and myriad forms—that the Constitution—through its philosophy of transformative equality—was always meant to address, and will need to address in the years to come.
There is yet another way in which automated identification systems implicate questions of transformative equality: ‘by freezing its targets in time, portraying them as aggregates of their most difficult choices’.20 For example, one of the interveners in the Aadhaar challenge was the Safai Karmachari Aandolan (SKA), an organization that represents manual scavengers across the country. While responding to Aadhaar being made mandatory for accessing rehabilitation schemes for manual scavengers, Bezwada Wilson, as the head of SKA, observed:
… as manual scavengers we want to come out of the identity and destroy it forever. We don’t want that identity again. Any marginalized community, any manual scavenging or vulnerable community wants to destroy its existing identity; that is our whole struggle. So, you are branding; even if I come out of this and get liberation also, but in your Aadhaar, my occupation, where I come from, everything will be there.21
While the government argued in Court that Aadhaar fulfilled the dream of ‘One Nation, One Identity’, for someone like Bezwada Wilson, a fixed identity was more a nightmare than a dream: because it would ensure the continued reification of a structurally inegalitarian society.
Technological systems, of course, are by no means limited to the State. They are now at the heart of the ‘private regimes of power’, which the constitutional value of fraternity is concerned with. In some ways, this was evident in the Aadhaar challenge itself. As far back as 1964, the journalist Vance Packard had presciently observed how, in the case of centralized data banks, ‘we all fall under the control of the machine’s managers, to some extent’.22 In a situation where basic entitlements had been made subject to successful biometric authentication, that ‘control’ lay with the individuals who were tasked with manning the Point of Service (POS) machine on which biometric authentication was done as a precondition to distributing the monthly supply of grain.
The government argued that by replacing the middleman (who would manually check your identity) with the POS machine, a major source of bribery and corruption in the system had been eliminated through a simple, technological fix. As oral arguments in court revealed, however, the system had created a new class of middlemen, whose power lay in their control of the biometric drawbridge to something as basic as food.
The issues, however, go significantly beyond a national biometric authentication system. With the long retreat of the State from the economic life of the country, many decisions that directly impact fundamental rights are now made by private entities, which wield an increasing amount of power. Take, for example, the issue of healthcare in a nation that has failed to guarantee universal coverage to all its citizens. That role will come to be performed by private insurance companies whose business model depends upon excluding those at risk of sickness (i.e., those who need healthcare the most). As a reliable method of predicting the possibility of certain kinds of diseases, DNA information is thus an excellent proxy for insurers to shape their policies around, and exclude the most vulnerable constituencies of the population from coverage or charge them sky-high premiums. This is a recognized risk: in the US, for example, insurance companies and employers are barred by law from basing hiring or enrolment decisions upon genetic information.23
In early 2018, for the first time in India, a single judge of the Delhi High Court handed down a little-discussed but hugely important judgement that articulated a right against genetic discrimination under Article 14 of the Constitution.24 Unsurprisingly, the judgement was delivered in the context of insurance contracts. In particular, it was considering a contract that excluded ‘genetic disorders’ from the scope of coverage. In its recognition of genetic heritage as protected grounds that would trigger heightened scrutiny under Article 14, Justice Pratibha Singh extended the transformative reasoning of Naz Foundation, which we have discussed in Chapter 2.
But she also did something even more important. She held that the right against genetic discrimination extended both to State and private insurance contracts. Why? Because there is inequality of bargaining power between the insurer and the citizen, which allowed the latter no real say in determining the scope or terms of the relationship. As a matter of constitutional law, this is probably not sufficient to attract Article 14. But it does serve as a crucial recognition that the intersection of private regimes of power, individual rights, and technology is something that transformative constitutionalism can no longer ignore. An articulation of constitutional fraternity that addresses this web of relationships, therefore, is increasingly urgent.
The most obvious and visible impact of the advance of technology, of course, is upon the third pillar of the trinity: liberty. In the Aadhaar challenge, for example, the petitioners led with an argument based on surveillance. They attempted to demonstrate how a centralized information database, tracking every individual transaction through the authentication requirement, would cast a chilling effect upon the exercise of fundamental freedoms. The State’s response, interestingly, was not to justify surveillance, but to argue that the Aadhaar Act provided enough legal and technological safeguards that would make surveillance impossible. Across the Bar, therefore, the impact of centralized databases, of tracking and recording the daily transactions that constituted the web of an individual’s life, and of aggregating information silos—basically, of the ‘collection, storage and sharing about the behaviour and choices of … people’25—was not in dispute. What was in dispute was whether the design of Aadhaar enabled or precluded surveillance.
In August 2018, however, Anand Venkatanarayanan reported that the National Crime Records Bureau was making an increasingly aggressive push for access to the Aadhaar database for law-enforcement purposes, which was barred by the Aadhaar Act as it then stood.26 The importance of this is evident from the fact that during the Supreme Court hearing, the petitioners had a
rgued that the fingerprinting of the entire population for the purposes of checking future identity fraud treated everyone as a potential suspect, and recreated the jurisdiction of suspicion that had been the hallmark of the Criminal Tribes Act.27
The government responded by arguing that fingerprinting under Aadhaar was merely for identification purposes, and not for criminal law. A ‘merging’ of the nationwide Aadhaar and restricted police fingerprint databases would, of course, dissolve this distinction, and ensure that now everyone was a potential criminal suspect. Venkatanarayanan went on to point out that notwithstanding the UIDAI’s denial, many Indian state governments, such as the state of Andhra Pradesh, had already ‘created a vast fully interlinked resident database that has merged the crime and civilian aspects’.28 The local ‘state hub’ had:
… information about all its residents, the GPS coordinates of their homes, medicines they use, food rations they eat, what they say about their chief minister on their social media accounts, their caste, bank accounts on which they receive their scholarships, pensions and their Aadhaar numbers.29
Meanwhile, in the neighbouring state of Telangana, ‘not only [were] the Aadhaar numbers of drunken drivers … seeded into the crime database, their family members’ Aadhaar numbers were also seeded’.30 In Telangana’s capital, Hyderabad, the police would regularly demand the Aadhaar number of anyone arrested, regardless of whether they were eventually convicted or not. Therefore, Venkatanarayanan concluded, ‘a vast database [i.e., Aadhaar] that allows other entities (including states to build their own parallel and bigger) databases with parallel biometrics, by design, is exactly the architecture that an all-encompassing surveillance state would need’.31
The Supreme Court’s Aadhaar judgement was limited to the central Aadhaar database. It thus leaves open a host of issues pertaining to the constitutionality of the state data hubs and, indeed, to the function creep, i.e., the merging of the civilian and criminal databases. From the Chennai cop who uses FaceTagr to non-consensually take a photograph of a ‘suspicious person’ and upload it to a database (which might, one day, be the Aadhaar database32) to the 360-degree view of citizens that is offered by state databases, the dangers of surveillance are both real and present. Future courts have a heavy responsibility to adjudicate the fundamental rights claims that will inevitably come before them. But to do that, as Eubanks points out, it is crucial to understand the difference between old surveillance (individualized targeting) and new surveillance:
… in new data-based surveillance, the target often emerges from the data. The targeting comes after the data collection, not before. Massive amounts of information are collected on a wide variety of individuals and groups. Then, data is mined, analysed, and searched in order to identify possible targets for more thorough scrutiny … If the old surveillance was an eye in the sky, the new surveillance is a spider in a digital web, testing each connected strand for suspicious vibrations.33
It was this argument—that rights are affected in the design of databases and not by their potential abuse in some imagined future—that the Supreme Court in the Aadhaar case failed to engage with. It was left to Justice Chandrachud, in dissent, to point out this simple fact. It remains to be seen whether a future court will engage with it.
That is because the issues will not go away. Let us briefly return to the question of DNA data banks. In the US, the police use DNA information to engage in familial searches. DNA from a crime scene can return a partial match with an existing profile in the database, which indicates that a family member of the profile-holder has the exact matching DNA. This becomes the basis of the police investigation. As is obvious, it brings in entire families within the scope of a criminal investigation (with all its attendant problems), purely on the basis of a possible relation with the accused.34
Would this be permitted by the Indian Constitution, in a nation with a long history of violent police abuse? Then, there is the issue of spatial privacy: we shed DNA everywhere we go, providing a detailed and accurate map of our movements. To what extent do we retain our privacy rights over this ‘abandoned DNA’ is a question that other jurisdictions have already begun to ask.35 And finally, but perhaps most crucially, there is that deceptive-sounding phrase—predictive policing. Based upon the idea that a crime can be prevented before it happens, predictive policing, put simply, profiles people’s past behaviour to predict how they are likely to behave in the future. Effectively, this is behind the Chennai Police’s use of FaceTagr to ascertain whether any past criminals are in an area, and to start investigating before anything has happened. The constitutional problems with such an approach are manifold. As Eubanks points out, ‘equity requires the ability to develop and evolve … [but] mathematical models, by their nature, are based on the past, and on the assumption that patterns will repeat’.36
And DNA has the potential to do predictive policing on steroids. It is already argued that DNA analysis can reveal genetic predispositions towards certain kinds of behaviour. On the one hand, this is halfway towards reviving the discredited nineteenth-century notions of criminals by birth, now with the added gloss of modern science. And the road from genetic predispositions to eugenics is shorter and easier to traverse than most people recognize. On the other hand, the temptation for law enforcement, and for a court that has always been extraordinarily sympathetic to the demands of law enforcement, to use it with ‘appropriate safeguards’ is immense.
It is here that transformative constitutionalism—as articulated in Selvi—is crucial. There are some things, some invasions of personal rights, some exercises of State power that the Constitution simply does not allow, no matter what benefits they may be scientifically proven to have. Indeed, as scholars note, the increasing accuracy of automated decision-making, especially statistical prediction, will make starker the ‘trade-off between fairness and utility’.37 The choices will become more controversial and more painful to make. Selvi, I suggest, lays down some excellent principles about where and how, under a transformative Constitution, we must hold the line.
Technological and information systems impact freedom in a second way: by serving as drawbridges (and often, as walls) between individuals and their basic entitlements. A large part of the arguments in the Aadhaar challenge turned upon the question of exclusion, i.e., cases of biometric authentication failure, where the machine’s determination of who you are is given precedence over who you say that you are. The State argued that, through administrative orders, it would ensure that in cases of authentication failure, other forms of identification would be accepted, and that nobody would be deprived of their rights simply because their fingerprints (or iris scans) failed to match.
The constitutional correctness of that specific argument aside, there is a deeper debate to be had about the role of algorithmic decision-making in determining identity for the purpose of accessing welfare rights. There is now an emerging consensus that an individual has the right not to be subjected to a decision ‘based solely on automated processing … which produces legal effects [that] significantly affect him or her’.38 In a famous 1970 judgement called Goldberg v. Kelly, the US Supreme Court held that a right to a full hearing existed before someone’s welfare benefits were terminated. In an age where welfare benefits have now been made dependent upon biometric authentication, the right to a fair hearing (presently not mandated under the Aadhaar Regulations, and, in fairness, not argued before the Court) in this and other contexts may have to be judicially read in as a compulsory requirement into a statute that makes automated processing a precondition for access to any right or benefit.
Even with a right to a fair hearing, however, automated processing shifts the burden of identification upon the individual.39 Having been rejected by the machine, the individual must now, using other means, demonstrate that she is who she claims to be, and not a ghost.40 In its own way, this resets some of the fundamental axioms of our legal and constitutional system:
‘… it is better that ten guilty men
go free than one innocent man goes to jail … then modernization flipped that on its head.’ Automated eligibility [is] based on the assumption that it is better for ten eligible applicants to be denied public benefits than for one ineligible person to receive them.41
This again impacts more than one pillar of the constitutional trinity. Critiquing a proposed Australian Bill introducing facial recognition technology (which will also probably be a feature of Aadhaar authentication by the time this book is in your hands), Cynthia Wong noted, ‘errors shift the burden onto individuals to show they are not who the system says they are, undermining the presumption of innocence … and this may disproportionately impact already vulnerable communities if the system misidentifies them at higher rates’.42
When does this burden become too onerous to attract constitutional scrutiny? The majority did not believe that the threshold had been crossed in Aadhaar (refusing to even examine the evidence placed on record), while Justuce Chandrachud, on the other hand, observed, ‘Basic entitlements in matters such as foodgrain can brook no error. To deny food is to lead a family to destitution, malnutrition and even death.’43 Obviously, this too is a question that will occur and recur in the years to come.
There is also a deeper issue here. In Chapter 8 of this book, we discussed how the shift from a culture of authority to a culture of justification is integral to the project of transformative constitutionalism. When decision-making shifts to automated processes, this dichotomy becomes even sharper because ‘decisions … are made not by humans or even human-legible rules, but by less scrutable mathematical techniques’.44 In technical terms, this has been labelled as creating a black-box society.45 At its root, it is simply a replication of the culture of authority, updated for the information age.
The Transformative Constitution Page 39